It is proposed that humans use mental models adaptively in coping with their social environment. They construct in imagination mental models of persons categories, situation types, and themselves, and they know whether they like or dislike what they have imagined. Behavior is often influenced as much or more by evaluation of these imagined exemplars as by the pushes and pulls of the immediate environment. Attitudes are evaluations of the mental models that we construct. We tend to behave positively or negatively toward entire categories of persons because we always imagine the same exemplar when we think of the category. Reality interacts with these category prototypes in that attitudes are more likely to guide social behavior when the target person matches the imagined prototype. Proposed studies test whether category exemplars play a part in attitudes on abstract issues, the effects of experience on prototype realism, and the possibility that attitude change is better accomplished by a direct attack on the imagined category exemplar than by trying to disconfirm concomitant beliefs. Mental models are also important in determining the cross-situational consistency of behavior, in that situations are rendered concrete by imagining the typical person who copes best with each type. Proposed studies test whether strongly evaluated situation exemplars elicit rigid or overly consistent behavior across situations, and whether persons who are typical of a social category are expected to be especially consistent in their behavior. Finally, it is proposed that the self is an attitude prototype derived from the overlapping views that others hold of oneself. This perspective generates proposed studies that test whether demonstrations of increased self-efficacy change self-attitudes more when performed for a prototype-holding audience, and whether powerful others might unknowingly inflict their views of us on others, and through them have a magnified effect on our self-attitudes. The long-term goal is to identify a set of integrative principles that can be applied to questions in five areas of research: attitude-behavior consistency, attitude perseverance, attitude change, the cross-situational consistency of behavior, and the self.